Annual Report & Research Highlights

1

Global Calibration of the GEOS-5 L-band Microwave Radiative Transfer Model over Non-Frozen Land Using SMOS Observations

Gabriëlle De Lannoy, Rolf Reichle, and Valentijn Pauwels (Monash University)

Assimilating low-frequency (1-10 GHz) passive

microwave observations into land surface

models should improve estimates of land surface

conditions and hence weather and climate

predictions. Global observations of brightness

temperatures (Tb) are available from the L-band

Soil Moisture Ocean Salinity (SMOS), and

similarly low frequency Tb observations are

expected from the future NASA Soil Moisture

Active Passive (SMAP) mission. In this study, a

microwave radiative transfer model (RTM) is

coupled to the Goddard Earth Observing System,

Version-5 (GEOS-5) Catchment land surface

model in preparation for the assimilation of

global Tb observations from SMOS and SMAP

as part of a radiance-based soil moisture

analysis.

Figure 1 shows differences between

climatological mean SMOS Tb observations and

Tb forecasts using three different literature-

based sets of RTM-parameters:

“Lit1” refers to RTM parameters that are

proposed for the future SMAP L2/3

radiometer retrieval products,

“Lit2” refers to RTM parameters collected

from literature studies using the L-band

Microwave Emission of the Biosphere

Model (L-MEB), the Land Surface

Microwave Emission Model (LSMEM), or

the Community Microwave Emission

Modelling Platform (CMEM), and

“Lit3” parameters are identical to the Lit2

parameters except for the microwave

roughness h, which is set to values used for

SMOS monitoring at the European Center

for Medium-Range Weather Forecasts

(ECMWF).

All three sets of literature-based RTM

parameters lead to substantial biases when used

in the GEOS-5 system, with Lit1 being too cold

by up to 50 K and Lit3 being too warm all over

the globe. Even though Lit2 estimates are nearly

Figure 1: Difference in Kelvin between one-year (1 July 2010 – 1 July 2011) mean values of horizontally polarized Tb at 42.5° incidence angle from GEOS-5 and SMOS observations for (a) Lit1, (b) Lit2, (c) Lit3, and (d) calibrated parameters. Within each subplot, “avg(|.|)” indicates the average absolute difference across the globe (excluding regions impacted by open water or radio-frequency interference that are shown in white).

Annual Report & Research Highlights

2

unbiased in the global average, there are still significant regional biases in the simulated Tbs. Such biases

would hamper the assimilation of Tb from SMOS or SMAP. Therefore, we calibrated the relevant

RTMparameters to obtain climatologically unbiased Tb estimates. After calibration, the Tb simulations

show a global absolute bias of 2.7 K, as shown in Figure 1d. It should be emphasized that an RMSE of

approximately 10 K remains, due to seasonal biases and short-term errors, which will be addressed in the

Tb data assimilation system.

The calibration minimized the difference between modelled and observed climatological means and

standard deviations, as well as the deviation of the optimized parameters from prior guesses (Lit1, Lit2,

and Lit3), at each individual location. Through a number of experiments, we determined that it is best to

simultaneously calibrate (i) the microwave roughness h, (ii) the vegetation opacity τ and (iii) the

scattering albedo ω.

Figure 2 shows the prior guesses (Lit1, Lit2, and Lit3) for the RTM parameters along with the optimized

parameter values after calibration, averaged by vegetation class. Both the microwave soil roughness h

and vegetation opacity τ are time dependent (because h depends on soil moisture and τ depends on the

leaf area index) and are therefore presented as temporal averages (denoted with <.>). The figure suggests

that the h is too low in Lit1 and too high in Lit3. After calibration, h values are closest to Lit2. The τ

estimates distinguish clearly between high and low vegetation. Finally, ω is increased over low

vegetation to reduce the vegetation effect in the simulated Tb.

Publication:

De Lannoy, G. J. M, R. H. Reichle, and V. R. N. Pauwels, 2013: Global Calibration of the GEOS-5 L-

band Microwave Radiative Transfer Model over Non-Frozen Land Using SMOS Observations. J.

Hydromet., (in press), doi:10.1175/JHM-D-12-092.1.

Figure 2: (a) Time-mean <h> (1 July 2010 – 1 July 2011), (b) time-mean <τ>, and (c) time-invariant ω; (Lit1, Lit2 and Lit3) before calibration, and (Cal) after calibration, spatially averaged by vegetation class. International Geosphere-Biosphere Programme (IGBP) vegetation classes are (ENF) Evergreen Needleleaf Forest, (EBF) Evergreen Broadleaf Forest, (DNF) Deciduous Needleleaf Forest, (DBF) Deciduous Broadleaf Forest, (MXF) Mixed Forest, (CSH) Closed Shrublands, (OSH) Open Shrublands, (WSV) Woody Savannas, (SAV) Savannas, (GRS) Grasslands, (CRP) Croplands, (CRN) Cropland and Natural Vegetation, and (BAR) Barren or Sparsely Vegetated. Thin gray lines for Cal indicate the spatial standard deviation within each vegetation class.